# VAKYN > VAKYN makes decision models. They answer typed questions about your documents in milliseconds, say how sure they are, and keep getting better at your work, on your servers or ours. Facts for writing code against VAKYN: - One endpoint: `POST https://api.vakyn.com/v1/systemone` with `Authorization: Bearer `. It is a Jev-compatible API, so the official TypeSafe SDKs (Python, JavaScript) work by changing the base URL and key. - A request has a `state` (text or JSON: the document) and named `questions`. Each question has a `type`: `noul` (yes or no; the answer is the probability of yes), `choice` (one of up to 255 named options, with a probability for each and a confidence) or `score` (a level on an ordered scale, with the expected level, a probability per level and a confidence). - The answer is one typed answer per question name, plus `usage.input_tokens` and `usage.output_tokens`. Nothing is generated, so there is no text to parse. Act on answers above your own confidence threshold and send the rest to a person or an LLM. - Price: input $30 per billion tokens; output tokens are free. Keys and prepaid balance: https://vakyn.com/console. Self-hosting: https://github.com/vakyn-ai/open-server. - Agent skills for building with VAKYN: `claude plugin marketplace add vakyn-ai/skills` then `claude plugin install vakyn@vakyn-ai` (Claude Code), `npx skills add vakyn-ai/skills --skill '*' -a codex -a cursor -a opencode` (other agents). ## Docs - [Overview](https://vakyn.com/docs): What VAKYN answers and where to start. - [Quickstart](https://vakyn.com/docs/quickstart): Get a VAKYN MAX key and make the first call. - [Concepts](https://vakyn.com/docs/concepts): The state, the questions, and what comes back. - [How it works](https://vakyn.com/docs/how-it-works): Pick a question type, then follow a request from state to decision. ## Questions - [Noul](https://vakyn.com/docs/noul): Yes or no, as the probability of yes. - [Choice](https://vakyn.com/docs/choice): Pick one of up to 255 named options. - [Score](https://vakyn.com/docs/score): Place the state on an ordered scale of up to 10 levels. - [Confidence](https://vakyn.com/docs/confidence): How sure an answer is, and how to use it. - [Patterns](https://vakyn.com/docs/patterns): Common shapes around one call: routing, fan-out, composite scores, guardrails. - [Routing by confidence](https://vakyn.com/docs/patterns/routing-by-confidence): Act on the sure answers, send the rest to a person or a larger model. - [Routing by intent](https://vakyn.com/docs/patterns/routing-by-intent): A choice question picks the handler, queue or team. - [Fan-out](https://vakyn.com/docs/patterns/fan-out): Many questions in one request, the state read once. - [Composite scores](https://vakyn.com/docs/patterns/composite-scores): Several scores combined into one number with your weights. - [Guardrails for agents](https://vakyn.com/docs/patterns/guardrails-for-agents): Check an agent's tool call before it runs. - [Long documents](https://vakyn.com/docs/patterns/long-documents): Split what does not fit and combine the answers. - [Many requests at once](https://vakyn.com/docs/patterns/concurrency): Send work concurrently and let the server queue it. ## Reference - [API reference](https://vakyn.com/docs/api): Endpoints, schemas, limits, prices, errors and headers. - [Python SDK](https://vakyn.com/docs/python): The official typesafe-sdk package, pointed at VAKYN. - [JavaScript SDK](https://vakyn.com/docs/javascript): The official @typesafe-ai/sdk package, pointed at VAKYN. - [Switching](https://vakyn.com/docs/switching): From a hosted Jev-compatible API: change the base URL and the key. - [Self-hosting](https://vakyn.com/docs/self-hosting): open-server: install, commands, flags, the model, backups and Docker. ## Skills - [/vakyn:start](https://vakyn.com/skills#vakyn): Set up. Checks whether VAKYN fits, picks a model file for your machine, sets up the server and makes a first call. - [/vakyn:research](https://vakyn.com/skills#vakyn-research): Research. Finds the decisions a chat model or a person makes in your app or process, and ranks them. - [/vakyn:questions](https://vakyn.com/skills#vakyn-questions): Questions. Picks noul, choice or score, writes the options and levels, and fixes a question that answers badly. - [/vakyn:convert](https://vakyn.com/skills#vakyn-convert): Convert. Turns a chat-model prompt, an output schema or a tool list into VAKYN calls, with Python or TypeScript code. - [/vakyn:workflow](https://vakyn.com/skills#vakyn-workflow): Workflow. Designs a new decision workflow from a task in plain words, such as routing, triage, picking a tool or checking a claim. - [/vakyn:cascade](https://vakyn.com/skills#vakyn-cascade): Cascade. A cheap chat model drafts, VAKYN checks the draft against its sources, and only failed drafts go to the expensive model or a person. - [/vakyn:judge](https://vakyn.com/skills#vakyn-judge): Judge. Grades chat-model or agent outputs against your rubric, lines the grades up with your experts, and runs in CI. - [/vakyn:batch](https://vakyn.com/skills#vakyn-batch): Batch. Runs a CSV, JSONL, Parquet or SQL table through VAKYN: a pilot first, then the full run with resume and progress. - [/vakyn:distill](https://vakyn.com/skills#vakyn-distill): Distill. Turns an agent that follows a written procedure into a program. Code does the exact steps, VAKYN questions make the decisions, small agent steps do look-ups, and a person takes the hard calls. - [/vakyn:migrate](https://vakyn.com/skills#vakyn-migrate): Migrate. Moves code that calls a hosted decision API to your own VAKYN server, and points framework adapters at it. - [/vakyn:check](https://vakyn.com/skills#vakyn-check): Check. Tests VAKYN against your chat model on about 30 of your records, reads side-by-side logs, sets thresholds, checks calibration and drift, and reviews the code before go-live. ## Optional - [All docs as Markdown](https://vakyn.com/llms-full.txt): every docs page in one file. - [FAQ](https://vakyn.com/faq): pricing, data, hosting and speed questions. - [How we measured](https://vakyn.com/how-we-measured): the speed and cost race against chat models. - [Lore](https://vakyn.com/lore): what VAKYN believes about decisions at volume. - [Blog](https://vakyn.com/blog): notes on deciding at volume. - [Contact](https://vakyn.com/contact): pilots on your own data, support, security.